Insight Recommender System Based on Text Network Analysis and GPT AI

InfraNodus has a built-in tool that analyzes the connections between your ideas and recommends the next interesting idea that you could have. This can be very interesting for research, creative writing, and data analysis.

It analyzes the connections you've already made between your existing ideas using the graph and GPT AI language models. It will then propose you to think of the connections you haven't thought of before. Once you add the connections, it will propose you to develop the peripheral ideas or to explore the less developed topics. InfraNodus will also make use of OpenAI's GPT natural language generator to ask you the research questions that would link those concepts together.

In this way InfraNodus follows an ecological model of discourse development, based on the assumption that it is important to both connect the existing ideas and to disrupt them in order to develop the discourse and make it diverse at the same time. Focusing on he specificies (zooming in) while also having an overview (zooming out) — as shown on the schema below.

You can use this tool in real time or with the texts that you've written before or imported from other sources (files, Evernote, Google, Twitter, newsfeeds, scientific literature etc).

 

See our Quick Tutorial on the Insight Recommender System:

How it Works? Technical Details

InfraNodus uses text network analysis to generate insight about any discourse or connected data. The basic appoach is to first identify the main clusters (based on the modularity and Louvain community detection algorithms) and the most influential nodes (based on the nodes' betweenness centrality). This will provide a good overview of a discourse or a data set, the main topical clusters contained within and the connections betwen them.

The next step is to identify the structural gaps between these clusters that could be bridged. These are shown both on the graph and in the analytics panel (Insight). InfraNodus will then propose you to make a connection between the two topical clusters that are not connected yet, bridging the gap to generate a new ideas.

These gaps are also referred to as the "structural holes" in the social sciences. The connections between them have a high potential for generating new ideas and innovation. As you are adding your ideas into the network graph, InfraNodus will recommend you the new connections and research questions that can help you develop your discourse and generate new innovative ideas based on this approach.

In addition to analyzing the graph, we use OpenAI's GPT powerful natural language generator model. It will use the structural gap data to generate relevant research questions and ideas for you.

Generating Ideas using InfraNodus and GPT:

This Video Explains the Technical Approach behind the Workflow in Detail:

Try It Yourself

You can try this approach yourself using InfraNodus for your own ideas, notes, and thoughts.

Sign Up

Comparative Text Analysis: Finding Your Niche

Another interesting approach is to compare the two texts in order to see how they differ: what is contained in one of them that is missing from the other.

Imagine you have a research topic you're interested in and you want to know better what aspect of the topic you should focus on.

Using our Market Niche / SEO app, you could visualize Google search results for the topic as one graph and related search queries as another. You will then see these two graphs superimposed, so you can quickly find what exists in the search queries (demand), which is not yet in the search results (supply).

This way you can quickly identify an attractive and not very competitive niche to focus your efforts on.

Alternatively, if you wanted to sync with the discourse (e.g. put something out there that will be hype and understood well, but maybe a bit less innovative), it would make more sense to focus on the most prominent topics, but making sure that you bring something unique to the discourse by focusing on the periphery, the missing parts, and the structural gaps.

Step-by-Step Guide: Finding a Niche Between Different Contexts

Try It Yourself

You can try this approach yourself using InfraNodus for your own ideas, notes, and thoughts.

Sign Up
 

 

Custom Service: Insight Generation

Please, let us know if you have an interesting proposition, use case, research project, or service request.

Contact Us